If with blurring you mean the time-symmetric ('acausal') shape of the impulse response of the 22.05kHz anti-aliasing filter that is to be used in the production of CDs, and if you accept that this blurring is a
real issue (and I my view there is not shred of evidence for this), then you have to look at Meridian's top-flight CD-player, the 808.2, to see how Bob Stuart tackles the problem there.
Here is the Stereophile test
https://www.stereophile.com/cdplaye...re_reference_cd_playerpreamplifier/index.html, but sadly JA omitted the more interesting things from his measurement regime. PM did better at Hifi News, but you have to register to access the data (June 2008):
http://www.milleraudioresearch.com/avtech/index.html.
So what did Meridian do to get rid of that pesky (or not) filter pre-ringing that is embedded in nearly all CDs?
They use a minimum-phase low-pass filter in their oversampler, cutting in at lower than 22kHz. Minimum-phase means that it has non-linear phase distortion, but also that it has no pre-ringing, only post-ringing. By cutting below 22kHz it removes the ringing artefacts of the 22kHz AA filter used in the majority of CDs' production, but it, of course, imposes its own, different, artefacts. Meridian named this an 'apodising' filter, borrowing from optics were a light path is more heavily attenuated when one moves to the rim of the lenses.
Anyone can do to their CD signals what Meridian did at the replay stage, and anyone can do this as part of the recording process or as part of the replay process. All one needs is a tool like the fully configurable resampler in iZotope, and these days probably a zillion other software packages. You set it for a high-order low-pass at 20kHz or so, minimum phase, and let rip.
Salient detail 1: with this style of filter Meridian did nothing else than emulating the late-70s/early-80s analogue anti-alias filter of digital recording systems like the Sony PCM-1630, the same filters that everyone rallied
against in those times. Moreover, Meridian were not first: one of the filters in the Marantz CD-7 did pretty much the same, only Marantz made no brouhaha about this.
Salient detail 2: Ayre reacted to Meridian's filter by finely pointing out that it, still, had very much post-ringing, proposing a much shorter filter, also minimum phase, and deploying that in their products. Ayre named that too 'apodising', not grasping that due to its shortness it would lack the stopband rejection to actually cut out the recording time ringing.
This to bring us back to the use of MQA at CD rates ...
First it must be made clear that MQA-for-CD-rate is a afterthought. It never was part of the original MQA idea, the spark that made Craven and Stuart devise a method for packing a high-resolution (i.e. 88.2kHz or more) signal into something more or less compliant with base resolution (i.e. 44.1kHz or 48kHz) PCM. So baseline CD is
not MQA's core. It was added to the MQA story, presumably in order to be able to infect more music, extract more money. After all the high-res market and music pool is very limited and shareholder value must be created.
When I ran my tests of MQA a couple of years ago, using Tidal, a Meridian Explorer2, and digital capture of raw MQA data as well as Tidal-unfolded data, it became clear that for baseline CD rate, MQA does
not use the Meridian 808.2 approach. What MQA does is much closer to the Ayre leaky filter style, i.e. clearly suboptimal in each and every respect. I theorised about the why of this back then, but as it is a long time ago I have forgotten much. Maybe I'll revisit my old files. Maybe not.
Now looking today at GO's baseline files I wonder what is going on. But it is hard to see, as GO's masters clearly overloaded the MQA encoder, yielding a veritable dog's breakfast.
